LLMBDC: Language Model for Biological Domains Oriented Clustering of Gene Ontology

2026-07-30

AI Summary

Gene Ontology (GO) enrichment analysis often struggles to distill high-level biological themes due to semantic redundancy among terms, and existing methods lack contextual awareness. To address this, this work proposes a training-free framework that leverages the zero-shot reasoning capability of large language models for GO term clustering—a first in the field. Relying solely on ontological information, the approach automatically generates interpretable biological domains (BioDomains) through semantic reasoning, confidence scoring, and Cauchy combination testing. Evaluated on Alzheimer’s disease and Fragile X syndrome datasets, the method substantially outperforms six baseline approaches in terms of Adjusted Rand Index (ARI) and Normalized Mutual Information (NMI), while effectively preserving statistically significant functional signals. This enables context-aware, scalable, and interpretable system-level functional annotation summarization.

Abstract

Gene Ontology (GO) enrichment analysis is a foundational tool for translating large-scale genomic data into biological insights, but typically yields hundreds of redundant terms that obscure overarching themes. Existing summarization tools rely on fixed similarity metrics (REVIGO, GOSemSim, clusterProfiler::simplify()), gene-overlap measures (Metascape), or static hierarchy mappings (GO-slim), and therefore cannot incorporate biological context. Manual curation provides context-aware grouping but is subjective and labor-intensive. A scalable, context-aware framework is needed to cluster GO terms into interpretable higher-order biological domains. Here we present LLMBDC (Large Language Model for Biological Domains Oriented Clustering of Gene Ontology), a training-free framework that leverages zero-shot semantic reasoning of LLMs with confidence scoring to cluster GO terms into BioDomains using only ontology information at inference time. Benchmarked across Alzheimer's disease (AD) and Fragile X syndrome (FXS) against six baseline methods including SapBERT, LLMBDC achieved substantially higher precision, recall, and clustering performance. Against ground-truth annotations, LLMBDC improved ARI from 9.7% to 73.3% (AD) and from 15.7% to 66.6% (FXS) over REVIGO, with corresponding NMI gains from 59.9% to 73.4% (AD) and 66.0% to 79.5% (FXS). A Cauchy combination test further confirmed that aggregated BioDomains retained statistically significant functional signals. LLMBDC provides a scalable, reproducible, and interpretable route to context-aware, system-level interpretation of GO enrichment results while preserving biological specificity.

Problem

• Gene Ontology
• enrichment analysis
• redundancy
• context-aware clustering
• biological interpretation

Innovation

• zero-shot semantic reasoning
• context-aware clustering
• Gene Ontology enrichment
• large language model (LLM)
• BioDomains